The proposed research will involve the acoustic analysis of tape-recorded speech. Longitudinal data on the development of speech will be derived by following a sample of five normal children from age three months to three and one half years. The data sample will be increased as proves necessary, by recording either additional children in specific age ranges, or adding more children to the longitudinal group. Formant frequencies, fundamental frequencies, center frequencies of noise bursts and other spectral data will be derived by means of sound spectrograms made with both normal and very wide bandwidth analyzing filters. Computer implemented Fourier and "analysis-by-synthesis" techniques will also be used. The tape recordings will make use of microphone placements that enhance the balance of breathy excitation and so facilitate the measurement of formant frequencies. The data will be used to test and develop theories that account for both the development of speech production, and the internal structure of human phonetic and phonologic ability. We shall test new theories that involve the presence of feature detectors and the maturation of innate mechanisms. The data will also address the question of the effects of the child's linguistic environment and will, moreover, provide a quantitative data base for the description of "normal" speech development that is an essential precursor for more effective diagnosis and treatment of retarded speech development.